remove libm.h includes when math.h and float.h are enough

src/math/fdim.c

-#include "libm.h"
+#include <math.h>
 
 double fdim(double x, double y)
 {

src/math/fdimf.c

-#include "libm.h"
+#include <math.h>
 
 float fdimf(float x, float y)
 {

src/math/fdiml.c

-#include "libm.h"
+#include <math.h>
+#include <float.h>
 
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long double fdiml(long double x, long double y)

src/math/fmax.c

-#include "libm.h"
+#include <math.h>
 
 double fmax(double x, double y)
 {

src/math/fmaxf.c

-#include "libm.h"
+#include <math.h>
 
 float fmaxf(float x, float y)
 {

src/math/fmaxl.c

-#include "libm.h"
+#include <math.h>
+#include <float.h>
 
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long double fmaxl(long double x, long double y)

src/math/fmin.c

-#include "libm.h"
+#include <math.h>
 
 double fmin(double x, double y)
 {

src/math/fminf.c

-#include "libm.h"
+#include <math.h>
 
 float fminf(float x, float y)
 {

src/math/fminl.c

-#include "libm.h"
+#include <math.h>
+#include <float.h>
 
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long double fminl(long double x, long double y)

src/math/ilogb.c

@@ -11,7 +11,6 @@ int ilogb(double x)
 		if (u.bits == 0)
 			return FP_ILOGB0;
 		/* subnormal x */
-		// FIXME: scale up subnormals with a *0x1p53 or find top set bit with a better method
 		for (e = -0x3ff; u.bits < (uint64_t)1<<63; e--, u.bits<<=1);
 		return e;
 	}

src/math/ldexp.c

-#include "libm.h"
+#include <math.h>
 
 double ldexp(double x, int n)
 {

src/math/ldexpf.c

-#include "libm.h"
+#include <math.h>
 
 float ldexpf(float x, int n)
 {

src/math/ldexpl.c

-#include "libm.h"
+#include <math.h>
 
 long double ldexpl(long double x, int n)
 {

src/math/llrintl.c

-#include "libm.h"
+#include <math.h>
+#include <float.h>
+
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long long llrintl(long double x)
 {

src/math/llroundl.c

-#include "libm.h"
+#include <math.h>
+#include <float.h>
+
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long long llroundl(long double x)
 {

src/math/lrintl.c

-#include "libm.h"
+#include <math.h>
+#include <float.h>
+
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long lrintl(long double x)
 {

src/math/lroundl.c

-#include "libm.h"
+#include <math.h>
+#include <float.h>
+
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long lroundl(long double x)
 {

src/math/nearbyint.c

 #include <fenv.h>
-#include "libm.h"
+#include <math.h>
 
 /*
 rint may raise inexact (and it should not alter the fenv otherwise)

src/math/nearbyintf.c

 #include <fenv.h>
-#include "libm.h"
+#include <math.h>
 
 float nearbyintf(float x) {
 	fenv_t e;

src/math/nearbyintl.c

-#include "libm.h"
+#include <math.h>
+#include <float.h>
+
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long double nearbyintl(long double x)
 {

src/math/nexttoward.c

@@ -11,6 +11,7 @@
  */
 
 #include "libm.h"
+
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 double nexttoward(double x, long double y)
 {

src/math/nexttowardl.c

-#include "libm.h"
+#include <math.h>
 
 long double nexttowardl(long double x, long double y)
 {

src/math/remainderl.c

-#include "libm.h"
+#include <math.h>
+#include <float.h>
 
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long double remainderl(long double x, long double y)

src/math/round.c

@@ -25,7 +25,7 @@
  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-#include "libm.h"
+#include <math.h>
 
 double round(double x)
 {

src/math/roundf.c

@@ -25,7 +25,7 @@
  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-#include "libm.h"
+#include <math.h>
 
 float roundf(float x)
 {

src/math/roundl.c

@@ -25,7 +25,9 @@
  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-#include "libm.h"
+#include <math.h>
+#include <float.h>
+
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long double roundl(long double x)
 {

src/math/scalb.c

@@ -15,7 +15,7 @@
  * should use scalbn() instead.
  */
 
-#include "libm.h"
+#include <math.h>
 
 double scalb(double x, double fn)
 {

src/math/scalbf.c

@@ -13,7 +13,7 @@
  * ====================================================
  */
 
-#include "libm.h"
+#include <math.h>
 
 float scalbf(float x, float fn)
 {

src/math/scalbln.c

 #include <limits.h>
-#include "libm.h"
+#include <math.h>
 
 double scalbln(double x, long n)
 {

src/math/scalblnf.c

 #include <limits.h>
-#include "libm.h"
+#include <math.h>
 
 float scalblnf(float x, long n)
 {

src/math/scalblnl.c

 #include <limits.h>
-#include "libm.h"
+#include <math.h>
+#include <float.h>
 
 #if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024
 long double scalblnl(long double x, long n)